Macromolecules 1998, 31, 1859-1865
1859
Solubility and Micellization Behavior of C60 Fullerenes with Two Well-Defined Polymer Arms Haruyuki Okamura, Nobuhiro Ide, Masahiko Minoda, Koichi Komatsu, and Takeshi Fukuda* Institute for Chemical Research, Kyoto University, Uji, Kyoto 611, Japan Received November 17, 1997; Revised Manuscript Received January 12, 1998
ABSTRACT: 1,4-Disubstituted, low-polydispersity C60 derivatives of the types C60-(PS)2, C60-(PVP)2, and C60-(PS-PVP)2, where PS is polystyrene, PVP is poly(p-vinylphenol), and PS-PVP is a diblock copolymer of this sequence, were prepared by applying the nitroxide-controlled free radical polymerization technique and studied with their solubility behaviors in some organic solvents. The first clear experimental evidence was obtained for the formation of multimolecular micelles of C60-bearing polymers under certain conditions: for example, light-scattering measurements showed that the C60-(PVP)2, C60-(PS-PVP)2, and C60-(PS)2 samples with a number-average molecular weight of a polymer arm roughly about 10 000 formed stable micelles in dilute tetrahydrofuran (THF) solution with association numbers of about 20, 6, and 1 (no micellization), respectively. The solvent power of THF for the mother polymers increases in this order. The saturation solubilities SC60 of the C60 moiety in C60-(PS)2 and C60-(PVP)2 were determined as a function of the PS and PVP chain lengths, showing that, in THF, the SC60 in the PS adduct is exceptionally large, much larger than that in the PVP adduct of the same chain length, in accord with the mentioned micellization tendency in dilute solution. On the other hand, C60-(PVP)2 showed a reasonable solubility in a polar solvent (methanol), in which C60-(PS)2 was little soluble. The micellization was found to be accompanied by characteristic changes in the UV-vis spectra, depending on micelle size.
Introduction Fullerene has attracted much attention due to its unique chemical and physical properties.1 Unfortunately, the ability to fabricate fullerene-based devices has been limited due to its poor solubility and processability. Polymer-bound fullerenes are particularly interesting because they can have the high solubility and processability of polymers as well as the unique properties of fullerene. Hence various kinds of C60-containing polymers have been synthesized and studied with their optical, electrical, and solubility properties.2-6 However, the information provided by these studies has often been limited. As indicated above, C60 has very limited solubilities in solvents: no solvent is known that is miscible with C60 in all proportions. The maximum saturation solubility has been obtained with 1-chloronaphthalene at 51 mg/mL.7 Although C60 has virtually no solubility in water, it can be solubilized in surfactant solutions via incorporation into the hydrophobic core.8 Interestingly, the formation of colloidal forms of C60 in aqueous surfactant solutions drastically changes the UV-vis spectrum from that in a molecularly dispersed state.8a Another useful method to solubilize C60 is derivatization. The solubility of a C60-containing polymer is, in a highmolecular weight limit of the polymer, expected to be essentially the same as that of the mother polymer, as in fact reported.4d However, the concentration of the C60 moiety itself in such a limit is small because of the small fraction of C60 in the polymer. The maximum saturation solubility of the C60 moiety in a solvent will be obtained when the C60 fraction in the polymer and the solubility of the mother polymer are optimized. Thus, the discussion of C60 solubility by derivatization requires us to study the solubility of the derivatives as a function of the chain length of the polymer moiety as
well as the solubility of the mother polymer itself. Such a systematic study has never been reported before this work. Here we have prepared 1,4-disubstituted, lowpolydispersity C60 derivatives of the types C60-(PS)2, C60-(PVP)2, and C60-(PS-PVP)2 by applying the nitroxide-controlled free radical polymerization technique,9 where PS, PVP, and PS-PVP denote polystyrene, poly(p-vinylphenol),10 and PS-PVP type diblock copolymer,10 respectively. This synthetic procedure was previously shown to be a simple and versatile one to yield well-defined C60 1,4-bisadducts.5 Another important problem addressed in this work is the states of solution, i.e., monomolecularly dispersed versus multimolecularly associated state of the C60polymers in solution. The first direct experimental evidence will be presented showing that the C60polymers form multimolecular micelles depending on the solubility and length of the polymer moiety and that the micellization accompanies characteristic changes in the UV-vis spectra, similar to those observed for pure C60 in aqueous surfactant solutions.8a Experimental Section Measurements. NMR spectra were observed at 270 MHz for 1H and 100 MHz for 13C NMR. UV-vis spectra were taken on a Shimadzu UV-2100. Static light-scattering measurements were made in tetrahydrofuran (THF) solvent at 25 °C by a DLS-7000 photometer (Otsuka Electronics, Japan), which was calibrated with benzene. The refractive index increment (dn/dc) in THF solution at 25 °C was measured by a DRM1030 differential refractometer (Otsuka Electronics). Gel permeation chromatography (GPC) was carried out in THF on a Tosoh HLC-802UR chromatograph (Tokyo, Japan) equipped with polystyrene gel columns (G2500H6 + G3000H6 + G4000H6; exclusion limit ) 1.0 × 106; 8.0 mm i.d. × 60 cm) and refractive index/ultraviolet dual-mode detectors. The system was calibrated with Tosoh standard PSs.
S0024-9297(97)01696-3 CCC: $15.00 © 1998 American Chemical Society Published on Web 02/24/1998
1860 Okamura et al.
Macromolecules, Vol. 31, No. 6, 1998 Scheme 1
Materials. C60, THF, o-dichlorobenzene (ODCB), benzoyl peroxide (BPO; Nacalai Tesque, Japan), and styrene were purified as described previously.2 p-tert-Butoxystyrene, kindly donated by Hokko Chemicals (Japan), was washed three times with 10 wt % aqueous NaOH solution and three times with distilled water and dried over anhydrous sodium sulfate. The filtered dry monomer was stored at -15 °C in an ampule.10 TEMPO (Aldrich) was used as received. Deionized water was distilled before use. Preparation of PS-TEMPO and PBOS-TEMPO. The polystyryl adduct with TEMPO (PS-TEMPO) was prepared as described previously.2,9d In a typical run to prepare a PBOS adduct with TEMPO,10 a mixture of BOS, BPO (1.02 × 10-1 mol L-1), and TEMPO (1.05 × 10-1 mol L-1) was charged in a glass tube, degassed with several freeze-thaw cycles, and sealed off under vacuum. Then, the mixture was heated at 95 °C for 3.5 h and at 125 °C for 3 h to yield a polymer. The polymer was recovered as a precipitate from a large excess of methanol, purified by reprecipitation with a chloroform (solvent)/ methanol (nonsolvent) system, and thoroughly dried (conversion: 50.5%), which, according to the PS-calibrated GPC, had a number-average molecular weight Mn of 2600 and a Mw/ Mn ratio of 1.12, where Mw is the weight-average molecular weight. Preparation of PBOS-PS-TEMPO (Scheme 1). A PBOS-TEMPO adduct (0.30 g, Mn ) 6100, Mw/Mn ) 1.08), obtained as above, was dissolved in styrene (0.92 mL) and heated at 125 °C for 2 h to obtain a PBOS-PS type block copolymer end-capped by TEMPO (PBOS-PS-TEMPO: 0.45 g, Mn ) 13500, Mw/Mn ) 1.15). Preparation and Purification of C60-(PS)2, C60(PBOS)2, and C60-(PS-PBOS)2. Preparation and purification of C60-(PS)2, C60-(PBOS)2, and C60-(PS-PBOS)2 were carried out as described previously.2 For example, PSTEMPO (6.90 × 10-4 mol L-1) and C60 (2.78 × 10-3 mol L-1) were dissolved in ODCB (5.00 mL), charged in a glass tube, degassed with several freeze-thaw cycles, and sealed off under vacuum. Then, the mixture was heated at 145 °C for 24 h. The product was a mixture of C60-PS adducts, unreacted PS, and unreacted C60. Since the solubility of C60 in THF is very low, virtually all (unreacted) C60 was removed as a precipitate by pouring the reaction mixture into THF (50 mL). The supernatant was dried by evaporation, to which 5.0 mL of benzene was added to dissolve the polymer. To this solution was slowly added 5.0 mL of methanol, and the precipitate was recovered by decantation. Since C60-PS adducts and PS (or PS-TEMPO) have largely different solubilities in organic solvents, this process was effective enough to separate C60PS (precipitate) from PS (in solution). In fact, it was confirmed that the supernatant contained no C60 derivatives and that
Scheme 2
the PS-TEMPO adduct was perfectly soluble in the benzene/ methanol mixture. All precipitate was carefully collected by centrifugation. The decanted solution was confirmed to contain only unreacted PS by 1H NMR, UV, and GPC. C60(PBOS)2 and C60-(PS-PBOS)2 were prepared and purified similarly. Hydrolysis of PBOS Moieties (Scheme 2). Hydrolysis of PBOS and PBOS segments were carried out as described previously:10,11 For example, 1 g of PBOS was dissolved in 1,4dioxane (20 mL), and hydrobromic acid (8.6 N, 1.3 mL) was added. The solution was then stirred magnetically at room temperature for 24 h and poured into water (200 mL). The precipitate was filtered off, again dissolved in 1,4-dioxane (15 mL), precipitated in hexane (180 mL), recovered by filtration, and finally freeze-dried twice from 1,4-dioxane to give poly(p-vinylphenol) (PVP) as a white powder with a quantitative yield. Saturation Solubilities of C60-(PS)2 and C60-(PVP)2. All operations were conducted in an air-conditioned room kept at 25 °C. A 40 mg quantity of each polymer sample was mixed with 0.1 mL of a test solvent in a centrifuge tube. After 1 min of sonication, the mixture was centrifuged for 20 min. The supernatant (10 µL, measured by microsyringe) was diluted to 100 or 500 times by the test solvent, and the concentration of the C60 moiety was determined by measuring the UV-vis absorption intensity at 440 nm and using the known molar absorption coefficient of C60-(BS)2 at 440 nm.5
Results and Discussion Synthesis and Characterization of C60-(PBOS)2 and C60-(PS-PBOS)2. From a synthetic viewpoint, this work is a simple extension of the previous one5 in which well-defined 1,4-bisadducts of C60 with PSs were prepared by the reaction of PS-TEMPO adducts with
Macromolecules, Vol. 31, No. 6, 1998
Solubility and Micellization Behavior of Fullerenes 1861
Table 1. Reaction of TEMPO Adducts with C60a precursor (TEMPO adduct)
product (C60 adduct)
code
Mn (Mw/Mn) by GPCb
yieldc (%)
Mn (Mw/Mn) by GPCb
Mn by UVd
Mn calcde
PBOS-TEMPO-1 PBOS-TEMPO-2 PBOS-TEMPO-3 PBOS-TEMPO-4 PBOS-PS-TEMPO-5f
890 (1.16) 2600 (1.12) 5200 (1.13) 9800 (1.14) 13500 (1.15)
31 44 52 42 46
1320 (1.31) 4200 (1.27) 9300 (1.21) 17300 (1.24) 24600 (1.27)
1800 4600 10200 18800 26100
2300 5700 10900 20100 27500
a In o-dichlorobenzene (145 °C, 24 h); [TEMPO adduct] ) 3.5 × 10-5 mol L-1 and [C ] ) 1.4 × 10-4 mol L-1 in all cases. b Calibrated 60 by standard PSs. c 100(wt of PBOS in C60-PBOS)/(wt of PBOS-TEMPO). d Calculated on the basis of the molar absorption coefficient of C60-(BS)2 of 5.01 × 103 mol-1 mL cm-1 (440 nm). e Calculated for bisadduct structure. f Mn,PBOS/Mn,PS ) 6100/7400 (by GPC).
Figure 2. GPC curves for C60-(PBOS)2-3 by UV (440 nm) and RI compared with that for PBOS-TEMPO-3 by RI. Figure 1. CO ) 4:1.
13C
NMR spectrum of C60-(PBOS)2-1 in CS2/(CD3)2-
C60 at a high temperature. The preparation and characterization of the PBOS-TEMPO and PS-PBOSTEMPO adducts used in this work were described in detail elsewhere.10 A possible mechanism by which the PS bisadducts are selectively produced was proposed previously,5 which should also apply to the PBOS and PS-PBOS systems. Table 1 summarizes the result of the reaction of the TEMPO adducts with C60. The yields (for definition, see footnote c to Table 1) of the purified products, which we designate as C60-(PBOS)2 and C60-(PS-PBOS)2, ranged from 30% to 50%, showing no clear dependence on the molecular weight of the precursor alkoxyamine. Figure 1 shows the 13C NMR spectrum of C60(PBOS)2-1, where the attached number shows the code number of the precursor alkoxyamine. In addition to the peaks derived from the PBOS moiety, the aromatic carbons derived from the C60 moiety and the ipso carbon in the BOS unit attached to C60 are observed between 140 and 160 ppm. Figure 2 shows the GPC curves of C60-(PBOS)2-3 recorded by different detectors. The PBOS moiety is detectable by both UV-270 nm and RI, but not by UV440 nm, while the C60 moiety is detectable by all UV440 nm, UV-270 nm, and RI. The two GPC curves of C60-(PBOS)2-3 are nearly identical with each other, which indicates a chemical uniformity of the polymer. The two curves retain a narrow polydispersity and commonly show an Mn value of 9300, which is about twice that of the precursor PBOS-TEMPO-3. Similar results were obtained for all other samples (Table 1). This indicates that the reaction of PBOS-TEMPO with C60 leads predominantly to a bisadduct, as in the case of PS-TEMPO. Figure 3 compares the UV-vis spectra from C60(BS)2 and C60-(PBOS)2-1 solutions, where BS is the
Figure 3. UV-vis spectra of (a) C60-(BS)2 and (b) C60(PBOS)2-1 in cyclohexane (6.63 × 10-2 g mL-1).
[2-(benzoyloxy)-1-phenyl]ethyl group.5 The two spectra are very similar to each other. This means that the polymer derivative is also a 1,4-bisadduct, like C60(BS)2 and C60-(PS)2.5 On the basis of the molar absorption coefficient of C60-(BS)2 at 440 nm, the Mn of the polymer derivatives was estimated (Table 1). Generally, the values of Mn by GPC and those by UV well agree with each other, and both are close to those calculated for the bisadduct structure. (The GPC value for the lowest molecular weight sample C60-(PBOS)2-1 is appreciably smaller than the UV and calculated values. This is ascribed to the C60 moiety contributing little to the hydrodynamic volume of the molecule. As the molecular weight of the PBOS moiety increases, this effect becomes less important.5) We thus conclude that the reaction of PBOS-TEMPO or PBOS-PS-TEMPO with C60 provides well-defined PBOS- or PBOS-PSdisubstituted 1,4-dihydro[60]fullerenes, selectively. Hydrolysis of C60-(PBOS)2 and C60-(PS-PBOS)2. The hydrolysis of the BOS units of the above-mentioned C60 derivatives gave PVP polymers of the types C60(PVP)2 and C60-(PS-PVP)2 (see the Experimental Section.) The 1H NMR spectrum of C60-(PVP)2-1 is shown in Figure 4. The peak at 1.3 ppm ascribed to the tert-
1862 Okamura et al.
Macromolecules, Vol. 31, No. 6, 1998
Figure 6. Zimm plot for C60-(PVP)2-4 in THF at 25.0 °C. Table 2. Values of Mw of C60 Experimentally Observed in THF at 25 °C and Those Calculated
Figure 4. 1H NMR spectra of (a) C60-(PBOS)2-1 in CS2/(CD3)2CO ) 4:1 and (b) C60-(PVP)2-1 in DMSO.
Figure 5. GPC curves of (a) C60-(PS)2 (Mn ) 10 000, Mw/Mn ) 1.14), (b) C60-(PS-PVP)2-5, and (c) C60-(PVP)2-4 in THF by RI detector. The broken lines are for the respective precursor TEMPO adducts.
butyl group in BOS disappeared and the hydroxyl proton peak at 3.1 ppm newly appeared, which suggests a complete hydrolysis of the BOS unit. This was confirmed also by a 13C NMR analysis. Solute Dispersion State Viewed by GPC. Figure 5 compares the GPC curves of C60-(PS)2 (Mn ) 10 000, Mw/Mn ) 1.14),5 C60-(PVP)2-4, and C60-(PS-PVP)2-5 along with those of their precursor alkoxyamines. All these C60 derivatives (as well as their precursors) have nearly the same molecular weight and a low polydispersity (Mw/Mn e 1.15), according to the PS-calibrated GPC study made in THF before removing the tert-butyl groups. This hydrolysis is unlikely to cause degradation of the main chain. Nevertheless, the GPC curves of the PVP-containing polymers are very different from those of the PS polymers in two main aspects, as Figure 5 shows. First, the elution time of PVP-TEMPO-4 is appreciably longer, and its elution profile is somewhat broader, than those of the PS-TEMPO counterpart. This suggests that PVP is less soluble and has a smaller hydrodynamic volume in THF than PS and/or that there is some attractive mean force between PVP and the stationary phase. This effect is still observable, though less markedly, for PVP-PS-TEMPO-5. Second and more notably, the elution curves for the PVP-containing
code
10-5Mw,exp
10-4Mw,calcd
Mw,exp/Mw,calcd
C60-(PVP)2-4 C60-(PS-PVP)2-5
3.0a
1.4c 2.4d
21 6.3
1.5b
a Determined by light scattering with dn/dc ) 0.138 mL/g. Determined by light scattering with dn/dc ) 0.172 mL/g. c Calculated for the bisadduct structure using the GPC value for PBOS-TEMPO-4 (Table 1). d Calculated for the bisadduct structure using the GPC value for PBOS-PS-TEMPO-5 (Table 1).
b
C60 derivatives are composed of multiple peaks and strangely skewed, indicating the formation of multimolecular micelles of various sizes. Judging from the GPC curves, the micelle size of C60-(PS-PVP)2-5 may appear to be larger than that of C60-(PVP)2-4. However, this cannot be concluded since the micellar size is generally a function of concentration, and the concentration of the solutes migrating through the GPC column is not uniquely defined. Moreover, the possible solute-gel interaction suggested above would make it difficult to judge micellar sizes by GPC profiles. In fact, light scattering shows that C60-(PVP)2-4 forms a larger micelle in THF than the other derivative does (see below). Micellization Behavior Studied by Light Scattering. Static light-scattering measurements were made for the THF solutions of C60-(PVP)2-4 and C60(PS-PVP)2-5. To obtain an idea about the sizes of the micelles possibly formed by these polymers, the scattered light intensities (I) from their solutions with a fixed concentration (0.5 wt %) were measured at a fixed angle (90°) and compared with the intensities from the solutions of their precursors PVP-TEMPO-4 and PVPPS-TEMPO-5, respectively, measured under the same conditions. The intensity ratios were found to be I[C60(PVP)2-4]/I[PVP-TEMPO-4] ) 15 and I[C60-(PS-PVP)25]/I[PVP-PS-TEMPO-5] ) 6 respectively. This indicates that micelles formed by several molecules of the C60-polymer exist in the THF solution in both cases, since the precursors can be assumed to be molecularly dispersed in THF. To obtain more quantitative information, light-scattering measurements were made for the C60-polymers at varying angles and concentrations. An example of the Zimm diagram is given in Figure 6. The obtained values of Mw are compared with the theoretical (unimolecular) values in Table 2. Clearly, the C60polymers form in THF micelles composed of about 21 and 6 molecules of C60-(PVP)2-4 and C60-(PS-PVP)25, respectively. Namely, the association number of the C60-polymer with the homo-PVP arms is larger than
Macromolecules, Vol. 31, No. 6, 1998
Figure 7. Schematic representation of multimolecular micelles formed by (a) C60-(PVP)2-4 and (b) C60-(PS-PVP)2-5 in THF at 25 °C.
that with the PS-PVP block-copolymer arms. The derivatives of the type C60-(PS)2 are molecularly dispersed in THF, according to the previous GPC study.5 Recalling that THF is a better solvent for PS than for PVP and that it will give an intermediate solubility to PVP-PS block copolymers, we may tentatively suggest that the poorer is the solvent toward the polymer moiety, the stronger is the association tendency of C60polymer,5 if compared at a common level of molecular weight. Since the C60 moiety is hardly soluble in THF, the micelles formed by a C60-polymer will have a core of the C60 moieties surrounded by the fringes of the polymer moieties, as illustrated in Figure 7. Such a core-fringes structure is somewhat similar to the one often observed for block-copolymer micelles formed in dilute solution with a selective solvent.12,13 It is also noted that the Zimm plot in Figure 6 seems quite normal, giving no indication of increasing association number with increasing concentration nor suggesting micelle dissociation at low concentrations. The straight concentration envelope with a positive slope (positive apparent second virial coefficient) implies that the micelles are stable in the studied range of concentration, 1 < 103C (g/mL) < 4. In a strict sense, the lightscattering values of Mw given above are apparent ones, since the C60-polymers, particularly C60-(PS-PVP)25, are not uniform in the refractive index increment (dn/ dc) along the chain, so that any distribution in the chain lengths of the PS and/or PVP moieties brings about a distribution of composition (hence dn/dc) among different molecules. This invalidates the simple light-scattering theory.14 The effect of composition heterogeneity, however, should be rather minor in the present systems, since the polymers are fairly narrow in chain length distribution. Moreover, multimolecular micellization will effectively average out the composition heterogeneity, if there is any.12a The slope of the angular envelope in Figure 6 suggests a micelle radius of gyration, Rg, on the order of 40 nm. This value appears to be somewhat too large for the micellar structure suggested in Figure 7 and may not be very reliable. A possible cause for the overestimation of Rg may be the difficulty in preparing perfectly dust-free solutions (because of the limited amount of the C60-polymer available). Nevertheless, the comparison of this Rg with the value of about 20 nm obtained for the C60-(PS-PVP)2-5 solutions, which were purified similarly, suggests that the polymer fringes in the C60-(PVP)2-4 micelles are appreciably more extended than those in the C60-(PS-
Solubility and Micellization Behavior of Fullerenes 1863
Figure 8. Saturation solubility SC60 of the C60 moiety in THF at 25 °C as a function of the Mn of the PS arm (dot-dash line). The dotted line shows eq 1.
PVP)2-5 micelles because of a higher steric hindrance in the former than in the latter, as illustrated in Figure 7. Saturation Solubilities of C60-(PS)2 and C60(PVP)2. As already noted, THF is a good solvent for PS and a nonsolvent for C60. Thus it is expected that a C60-(PS)2 adduct with PS chains shorter than a critical length will have a limited solubility in THF depending on the PS chain length. Adducts with PS chains longer than the critical length will be miscible with THF in all proportions. In fact, the adduct with a molecular weight of the PS chain, Mn,PS, of about 1000 showed a limited miscibility with THF, the saturation solubility being 190 mg/mL of solution. On the other hand, the adduct with Mn,PS = 2400 seems to be soluble in THF at all concentrations. In view of the generally poor solubility of C60 in solvents and polymers, more interesting is the saturation solubility SC60 of the C60 moiety rather than that of the adduct as a whole. When the chain length of the PS moiety increases, the concentration of the C60 moiety within the bulk adduct decreases according to
SC60 ) (MC60/Madduct)dadduct
(1)
Here, MC60 and Madduct are the (number-average) molecular weights of the C60 moiety and the whole adduct, respectively, and dadduct is the adduct density, which we approximate here by the PS density of 1.06 g/mL. Equation 1 is shown by the dotted line in Figure 8, and it gives the maximum possible values of SC60 in any solvent. When a C60-polymer adduct shows a limited solubility in a given solvent, its value of SC60 is necessarily smaller than the dotted line. Therefore, the SC60 of the C60-(PS)2/THF system should behave like the dot-dash line in Figure 8, which is schematic to some extent because of the lack of a sufficient number of data points. All the C60-(PVP)2 samples prepared in this work showed a limited solubility in THF. Figure 9 gives SC60 as a function of the molecular weight of the PVP chain, Mn,PVP. SC60 is considerably smaller than the maximum possible value for the bulk adduct and seems to have a peak at around Mn,PVP ) 1000, where about 16 mg of the C60 moiety can be solubilized in 1 mL of THF solution. This maximum value of SC60 is considerably smaller than the one attainable by the PS adduct (Figure 8). This is ascribed to the poorer solubility of PVP in THF than that of PS and is consistent with the micellization tendency described in previous sections.
1864 Okamura et al.
Figure 9. Saturation solubility SC60 of the C60 moiety in THF (dot-dash line) and in methanol (solid line) at 25 °C as a function of the Mn of the attached PVP chain.
In this connection, it should be remembered that both the GPC and light scattering studies on the micellization behavior were made at very low concentrations. For example, the highest concentration of C60-(PVP)2-4 studied by light scattering (Figure 6) is 4.0 mg/mL in the adduct concentration or about 0.04 mg/mL in the C60-moiety concentration, which, of course, is far below the saturation-solubility line in Figure 9. Since micelles formed at low concentrations are unlikely to dissociate to single molecules at higher concentrations, it is indicated that C60-(PVP)2 adducts are solubilized in THF in a multimolecular micellar form in virtually the whole region below the SC60 line. On the other hand, all available C60-(PS)2 adducts (Mn,PS g 1000) are molecularly dispersed in dilute THF solution, as is judged from their GPC curves.5 It is an interesting open question whether they remain in a molecularly dispersed state when the concentration is increased toward the SC60 line in Figure 8. This question is somehow related to the likewise interesting and important one as to the molecular dispersion state in bulk C60polymers. Another interesting topic is the solubility of C60(PVP)2 adducts in polar solvents. We have determined the saturation solubilities of the PVP adducts in methanol, which seemingly is a good solvent for PVP. As in THF, however, all the C60-(PVP)2 adducts showed a limited solubility in methanol. Figure 9 gives SC60 as a function of Mn,PVP. This SC60 curve is similar in shape and magnitude to that in THF and suggests that methanol is not a strong enough solvent to disperse the adduct monomolecularly: namely, it is highly likely that the molecules are dissolved in methanol, forming multimolecular micelles again. At this time, it is not possible to confirm this by light scattering due to the technical difficulty of conducting measurements in methanol. In any case, it should be stressed that the C60 moiety can be made soluble in the polar solvent in concentrations over 1 wt % by the PVP derivatization. Micellization Behavior Studied by UV-Vis Spectra. The electronic properties of C60 are affected by its environment. Estoe8a reported that the UV-vis spectrum of the C60 molecules solubilized in water in a colloidal form by use of a surfactant is markedly different from that of the molecules in a monomeric dispersion state. Figure 10 shows the UV-vis spectra of three derivatives of nearly the same size (in THF): C60-(PS)2 (solid line; Mn,PS ) 5500, Mw/Mn ) 1.18), C60(PS-PVP)2-5 (broken line), and C60-(PVP)2-4 (dotted line). C60-(PS)2 shows the spectrum characteristic of
Macromolecules, Vol. 31, No. 6, 1998
Figure 10. UV-vis spectra of C60-(PS)2 (solid line; Mn ) 5500, Mw/Mn ) 1.18), C60-(PS-PVP)2-5 (broken line), and C60(PVP)2-4 (dotted line) in THF.
1,4-disubstituted C60, while C60-(PVP)2 gives a spectrum with no absorption maxima, which is rather similar to the one observed for the pure C60 in water in a colloidal form.8a On the other hand, the 440 nm peak of C60-(PS-PVP)2 is somewhat broadened and appears between the spectra of C60-(PS)2 and C60-(PVP)2. Remembering that the association numbers of the PVP, PS-PVP, and PS derivatives observed in this solvent are 21, 6, and 1 (no association), respectively, and that the colloidal micelles of C60 observed in water are of a macroscopic size, we may conclude that UV-vis spectroscopy offers a quick and convenient means to study the dispersion state of C60 compounds. It can even inform us of the approximate micellar size, if the association number is not too large, e.g.,
